AI fast-forwards video for sports highlights

AI innovation and hyper-personalized user experiences let broadcasters engage with viewers like never before

Imagine the manual workload broadcasters undertake to compile ten-minute recaps of the dozens of live sporting events taking place daily all over the world. Then, factor in what the human eye misses. People can’t possibly catch every play, pitch, goal, fumble, or flag.

Highlights, whether sports, film, or television, are the fastest growing segment of video, with the enterprise video market expected to grow to nearly USD 20 billion by 2023. Aiming to grab more of that burgeoning market, content creators can use artificial intelligence (AI) to analyze massive amounts of video and data.

Keeping an eye on the ball

At events the magnitude of the Wimbledon Tennis Championship or FIFA World Cup, AI-enabled video can deliver stats and results faster than a production crew by leaps and bounds. Wimbledon installed an AI system in 2017, delivering highlights 15 minutes faster and resulting in 14.4 million views of video content with no human intervention.

AI gathers and analyzes data courtside, taking cues from an ace at 100 mph, spotting reactions by spectators, and using visual action recognition to work out which moments to keep for the highlights reel and which to discard.

AI speeds and simplifies the video highlights process using:

Point selection. To weigh which plays to include in a highlights video, visual recognition APIs rank signs of celebration, such as fist pumps or arms raised in victory after a winning a point.

Clipping and fine—tuning. A time limit is applied to clip length based on data analytics, including time stamps when points start and end. Visual recognition can also detect content that isn’t play, such as wide shots of the crowd, to be edited out.

Production. Story telling graphics and watermarks are added before clips are merged in a reusable industry format and handed off to production.

Distribution. Once content is approved, it’s distributed to digital editors for publishing and websites, mobile apps, and social media.

Breakthrough fan and market engagement

At the 2018 FIFA World Cup, FOX Sports launched an AI—based platform for fans to create and share their own customized soccer highlight videos. Fans could browse and filter FIFA’s archive of past and current matches by year, team, player, and type of gameplay, such as penalty kicks and goals. The interactive platform then analyzed and compiled footage in seconds, at the user’s command. To keep post-match conversations going, fans could title, save, and share their creations across email, text, Facebook, and Twitter.

Creating personalized sports highlights according to one’s preferences can mean curating content based on a tournament, a team, favorite players, or certain plays. Other dimensions of preferences might be considered, like time spent watching and the type of devices preferred by the viewer, whether a five-inch handheld or a large screen. Preferences can also be dynamically discovered based on viewing behavior of pausing, replaying, and fast-forwarding content.

Platforms for engagement

With the ability to have new conversations with its audience, broadcasters become a platform for fan and market engagement. Advertisers will take notice, as will telecom companies investing in sport rights to differentiate “quad play” bundled packages of broadband, landline, mobile phone, and TV contracts to increase customer loyalty and average revenue per user (ARPU).

For over-the-top (OTT) content provided over a high-speed internet connection, innovation enabled by AI can help grow and retain clients. Auxiliary engagement channels, such as widgets or mobile apps, can drive ad inventory by creating relevant, contextual fan experiences that resonate with an advertisers’ message. In an ad-funded sports model, this experience creates ad slots that can be monetized.

Expanding the value and performance of video

Beyond sports, AI can help drive consumption of digital content by making it easier to classify and find. Whether from a subscription streaming service, such as Netflix, or an-ad funded model, the manner in which content is presented matters to consumers and how they choose to engage with it—or not.

Read the report to learn more about video experiences driven by data.


Bookmark this report  

Meet the authors

Mario Cavestany

Connect with author:

, VP, IBM Telecommunications, Media and Entertainment, (TME) Industries, Europe, IBM Global Markets

Fabien Lanne

Connect with author:

, Technical Leader, TME Industries, Europe, IBM Global Markets

Jay (Mrutyunjaya) Hiremath

Connect with author:

, Industry Solutions Leader, Media and Entertainment, Lead Partner, Global TME, IBM Consulting

Rob van den Dam, Global TM&E Industry Leader - IBM Institute for Business Value

Download report translations

Originally published 05 April 2019